Describes the methodological problems posed by sampling characteristics of nonorganic failure-to-thrive (NOFT) children and strategies to address them. Sources of variance in sample characteristics include the criteria used to define NOFT, populations from which samples are drawn, parent refusal and sample attrition, medical and psychological treatment, and individual differences in environmental or biologic risk factors. Undesirable consequences of unrecognized intrasample variation in NOFT include sample bias, limited generalizability of findings across different settings, and erroneous assumption of sample homogeneity. Comprehensive description of sample selection and characteristics will enhance more accurate definition of NOFT, generalizability of findings, and evaluation of the impact of sampling characteristics. Future studies should focus on objective assessment of subtypes of NOFT and the relationship of individual difference variables to psychological status and prognosis.